Business

Know the Business: Tencent Music Entertainment Group

TME is a high-margin subscription compounder wrapped in a regulated, melting live-streaming ice cube, sitting inside a holding-company shell that owns roughly ¥38B of net cash and securities. The market correctly recognises that consolidated revenue grew only ~16% in FY2025 — but the part the market is underestimating is that the surviving piece (online music) is a 40%+ operating-margin business with pricing power that has barely been tested. The part it may be overestimating is how quickly social entertainment will stop shrinking.

1. How This Business Actually Works

TME runs two engines glued together by one user graph. The subscription engine is a fixed-cost streaming utility: license a catalog from Universal/Sony/Warner plus domestic labels, host it on QQ Music / Kugou / Kuwo, charge ¥8–¥40 per month, and let the gross margin drop through. The social-entertainment engine is a transactional impulse-purchase business: users buy virtual gifts inside live-stream rooms or WeSing karaoke, and TME revenue-shares a large slice with the performer. The two share the same Weixin/QQ login funnel, the same recommendation infrastructure, and the same content licenses — but they have nothing in common economically.

The simple analogy. The subscription business is a toll bridge over Chinese music — TME pays the labels a relatively fixed annual "build cost" for the bridge (royalty minimums + revenue share), then collects a small toll from each user crossing. The toll is rising (ARPPU ¥10.0 → ¥11.8 over two years) and there is no limit, in principle, on how much more it can rise as the bridge improves (SVIP features, Dolby Atmos, K-pop early access). The live-streaming business is something else entirely: a flea market the platform rents out, where the platform takes a cut of every transaction but does not control the merchants and is regularly raided by inspectors.

What truly drives incremental profit. The royalty bucket (paid to labels + revenue-shared with live-streaming performers) collapsed from 78.9% of COGS in FY2023 to 61.8% in FY2025. Two forces did it. First, the 2021 antitrust ruling ended exclusive licensing in China — by FY2024–2025, contract renewals (Warner 2025, Sony 2025, Bin-music 2025) repriced down to normalised global royalty rates rather than the inflated exclusivity premiums of 2018–2021. Second, the mix shift away from social entertainment mechanically removed the highest-pass-through revenue line — live-streaming dollars came with 50–70%-of-revenue performer payouts; subscription dollars come with low-double-digit incremental content cost. Operating margin expanded from 21.8% (FY2023) to 40.6% (FY2025) on basically flat-to-modest revenue growth. That is the single most important fact about this business.

Bottlenecks and bargaining power. Upstream, the labels still set the cost floor — TME cannot run without Universal, Sony, Warner, and a handful of Chinese majors. But because exclusive licensing is now banned, NetEase Cloud Music can also access the same catalog at similar terms; the labels lose the bidding war that used to set the price. Downstream, the bottleneck is time-on-app, not content. Short-video apps (Douyin, Kuaishou, ByteDance's Qishui/Soda Music) compete for the same attention hour without paying the same licensing cost. Internally, the bottleneck is SVIP conversion — every standard subscriber pulled up to the ¥40+ tier roughly triples ARPPU contribution. TME ended 2025 with over 20M SVIP subscribers, a number that compounded at 50%+ YoY.

2. The Playing Field

TME shares no peer that does what it does. Spotify is the cleanest global on-demand-streaming benchmark but operates in a completely different competitive structure (no Tencent ecosystem, no virtual-gift economy, but a paying ratio many times higher). NetEase Cloud Music is the only same-game, same-country competitor but is roughly one-quarter the size. JOYY/HUYA are pure live-streaming comparables for the shrinking half of TME. Bilibili overlaps on the China interactive-media user base. The honest peer set is heterogeneous and that is the point — TME's economics look unlike any single comparable.

Note: Spotify, JOYY revenue/profit converted to CNY-equivalent at period-end rates for comparability. Spotify revenue 130,850 ≈ €17,186M; JOYY 15,080 ≈ US$2,124M.

What the peer set reveals. Three facts that should land harder than the table itself.

First, TME's operating margin (40.6%) is more than 3× Spotify's (12.8%) on a smaller absolute revenue base. That is structural, not transient — TME pays Chinese-market royalty rates post-antitrust, has a higher-margin advertising business inside the subscription product, and benefits from the WeSing/karaoke complement that generates content for free. The market has historically anchored TME's multiple to Spotify's growth profile while ignoring the margin advantage; the result is TME trading at 11.7× EV/EBITDA versus Spotify at ~33×.

Second, NetEase Cloud Music is now profitable and growing operating profit ~39% YoY, even with revenue ~one-quarter of TME. The cleanest read is that the duopoly is rationalising: both sides are pulling away from money-losing live-streaming and focusing on subscription economics. That removes the "Cloud Music is going to undercut TME on subs price" thesis some bears held, but it tightens the upstream — both players renew label contracts in a market where they cannot defect.

Third, the live-streaming peers (JOYY, HUYA) are economic warnings, not benchmarks: JOYY is profitable only because of a US$2.1B one-time sale gain; HUYA is loss-making and shrinking. They tell you what would have happened to TME if social entertainment had stayed the dominant engine.

3. Is This Business Cyclical?

It is cyclical in the social-entertainment half — which is now under one-fifth of revenue — and structurally compounding in the music-subscription half. The cycle that matters today is regulatory, not macroeconomic.

The margin chart tells the cycle story better than any prose. Gross margin compressed from 38.3% (FY18) to 30.1% (FY21) during the exclusive-licensing royalty inflation, then expanded to 44.2% (FY25) as exclusivity ended and the social-entertainment mix shrank. Operating margin compressed to 12.2% in FY21 and recovered to 40.6% in FY25 — a 28-percentage-point swing on roughly flat revenue. This was a cycle the income statement lived through, even if revenue growth makes it look like a steady-state business.

Where the cycle still hits today:

• Macro consumer spending. Live-stream gifting is pure discretionary spend and tracks Chinese tier-1 consumption sentiment with a short lag. The 2023–2025 China consumption softness shows up in the social-entertainment line, not the music line.
• Regulatory mood. Any new NRTA/CAC circular on virtual gifting can shave another billion ¥ off the social-entertainment line in a single quarter. This is now a tail risk, not a base case — the segment is already small.
• Label contract renewal windows. Warner (2025) and Sony (2025) renewals were absorbed at favourable terms. The next bunch (Universal due 2026–2027, plus various K-pop labels) is the next pricing test.
• Working capital and cash conversion. Days-payable-outstanding is 130+ days (¥6.3B accounts payable on cost of revenue, partly minimum-guarantee royalty accruals), so a label renegotiation that shortens payment terms would compress operating cash flow without hurting the income statement.

What is not cyclical here: utilisation, inventory, fixed-asset turnover. Capex is under 1% of revenue (¥305M on ¥32.9B FY25). There is no plant to run hot or cold; the cycle expresses itself in mix and margin.

4. The Metrics That Actually Matter

There are five operating metrics — three quantity, two price/mix — that explain almost all of TME's value creation. Most of the ratios investors instinctively reach for (revenue growth, MAU, EBITDA margin) either miss the point or duplicate one of these. One caveat for FY26 and beyond: management announced on the Q4 FY25 call that quarterly disclosure of paying users and monthly ARPPU will end — these will be reported annually only. The metrics themselves remain the right ones to track; the disclosure cadence is changing.

What the scorecard says. Every operating metric that drives value is improving; the ratios used to dismiss "China tech" stocks (slow revenue growth, declining MAU) misread what is happening here. Revenue grew only 16% in FY25, but operating profit grew 53% and FCF held above ¥9.9B for a second consecutive year despite ¥2B of buyback and dividend outflows — the signature of a late-phase structural margin transition, not stagnation.

5. What Is This Business Worth?

Value here is determined by online-music earnings power, not by consolidated revenue. The right lens is a sum-of-the-economic-parts, because the consolidated income statement blends two businesses with completely different multiples plus a financial-asset book that has nothing to do with either.

Why sum-of-parts, not a single multiple. Three reasons specific to this company.

First, the operating business and the financial-asset book do not belong on the same multiple. ¥38B of cash and short-term investments earn money-market-fund returns; a single P/E on consolidated net income misvalues both. The proof: FY25 net income includes a ¥2.4B one-time gain on deemed disposal (the Universal consortium distribution-in-kind) — strip it, and operating earnings power is much closer to the non-IFRS ¥9.9B figure.

Second, the online-music engine deserves a streaming/SaaS-like multiple (compounding, recurring, expanding margins). The social-entertainment engine deserves a melting-asset multiple — book value of the customer-facing IP and brand, recovered through residual cash flow. Mixing them produces a number that fits neither.

Third, the Tencent ecosystem dependency cuts both ways. Tencent owns 50%+ of equity and 90%+ of voting power. This is a moat (free distribution via QQ/Weixin, content-bundling power) and an overhang (capital-allocation control, related-party transactions, no chance of a takeover premium). Apply the same SOTP that you would to any Tencent-controlled affiliate (Huya, others), with the holding-company governance discount baked in.

6. What I'd Tell a Young Analyst

Stop modelling consolidated revenue growth. Model the segment mix. Consolidated revenue grew 16% in FY25; online-music revenue grew 23%; subscription revenue grew 16%; "other music services" grew 39%; social entertainment fell 7%. Each of those numbers tells a different story about what the business is becoming. The single biggest analyst mistake on TME has been treating it as "the Chinese Spotify" and benchmarking it on Spotify's growth profile — that ignores the much higher margins, much lower paying ratio, much lower ARPPU base, and the entire melting-ice-cube half of the business.

Watch four things every quarter, four things every year. Every quarter (until management's disclosure change takes effect): online-music subscription revenue, social-entertainment revenue trend, the service-cost-to-revenue ratio, and any NRTA/CAC circulars on virtual gifting. Every year (once new disclosure regime kicks in): total paying users at year-end, SVIP subscriber count, "other music services" revenue, and the cash + investments balance. If subscription revenue growth holds in the high teens, service-cost intensity stays under 35%, and SVIP keeps compounding, the thesis is intact regardless of what consolidated revenue does.

The market is most likely underestimating two things. First, that the operating-margin expansion from FY23 (21.8%) to FY25 (40.6%) is mostly structural — exclusivity-licensing royalty inflation is gone, and the mix shift toward higher-margin subscription is roughly two-thirds complete. Second, that ¥38B of cash plus financial-asset stakes plus an in-progress Ximalaya deal are not embedded in the consolidated P/E and should be valued separately. The market is most likely overestimating the speed at which Qishui/Soda Music substitution will hit TME's paying base — short-video music discovery and music subscription are partially complementary, not substitute, demand.

What would genuinely change the thesis. A label renewal cycle that reprices royalty rates back up by 200+ bps of revenue; a credible third Chinese music streaming entrant taking meaningful subscription share (so far ByteDance has user share via Douyin but not paying-subscription share); or a regulatory move that treats music subscription like the live-streaming crackdown treated virtual gifting. Absent these, value compounds with SVIP penetration, non-subscription IP monetisation, and disciplined capital return — and the market's biggest mistake is anchoring the whole thing to a number called "MAU."